I have no doubt that it has reasons for not providing the right answer, but the issue is that it didn't comprehend the question sufficiently to give the right answer.If you question it, it will say 3 but you have to pick up that it is going by pronunciation and then say, "I'm not talking about pronunciation but how many letters are actually in the word". Before that it will say 2, the first one and then the double RR![]()
ChatGPT didn't comprehend the question sufficiently whereas DeepSeek did (it even, per Q's post, showed its workings.)
I.e. chatGPT has issues with understanding that "how many Rs are there in STRAWBERRY?" is a request to know how many times the letter R appears in the word STRAWBERRY.
DeepSeek doesn't seem to have that issue.
Simples.